Resilience of the Pitt Bacteremia Score: 3 Decades and Counting

bloodstream infections, sepsis, mortality, qPitt, acute severity of illness

(See the Major Article by Henderson et al on pages 1826–33.)

Since its initial development and subsequent modifications by visionaries at the University of Pittsburgh, the Pitt bacteremia score (PBS) has been used for 3 decades to measure acute severity of illness and predict mortality in patients with bloodstream infections (BSIs) [1–5]. It has been utilized as a stratification tool in pivotal multicenter studies of BSIs due to Gram-positive bacteria, Gram-negative bacteria, and Candida species [6–8]. Patients with a PBS <4 have been considered noncritically ill, whereas those with a PBS ≥4 have been classified as critically ill due to a higher mortality risk. The PBS has outperformed other, more complex, commonly used acute severity of illness scores in a variety of clinical settings. For example, it had a higher discrimination for predicting mortality than did the Acute Physiology and Chronic Health Evaluation II (APACHE II) and the CURB-65 scoring systems in patients with sepsis in intensive care units (ICUs) and in Streptococcus pneumoniae BSIs, respectively [9, 10].

The simplicity of the PBS has been another major advantage over other commonly utilized scores. Whereas the APACHE (II-IV) and sepsis-related organ failure assessment (SOFA) scores include laboratory variables (serum creatinine, bilirubin, platelets, partial arterial oxygen pressure, etc), the PBS assesses acute severity of illness and identifies patients with life-threatening infections solely based on patient-specific variables that are determined during an initial physical examination. The absence of laboratory results allows for the immediate application of the PBS at the bedside, without delays for venous puncture and the subsequent receipt of laboratory results. Moreover, arterial blood gas measurements are not included in the PBS, which makes it more readily applicable among non–ICU housed patients than comparator scores.

The PBS is the cornerstone of the Bloodstream Infection Mortality Risk Score (BSIMRS), which has been derived and validated in patients with Gram-negative BSIs [11, 12]. The BSIMRS adds major comorbidities (cancer and liver cirrhosis) and the source of infection to the PBS to estimate the mortality associated with a BSI. The stratification of patients, based on their predicted prognoses at initial presentations, is critical in defining outcome expectations and precise estimations of the impacts of appropriate antimicrobial therapy on survival rates and hospital lengths of stay [13, 14]. Moreover, the BSIMRS has been used as a matching tool in observational cohort studies to allow for the evaluation of different management strategies in patients with comparable predicted mortality at the time of an intervention [15, 16].

Fever remains the Achilles’ heel of the PBS. The current (1998) version of the PBS allocates 1 point for a temperature of 39.0–39.9°C and 2 points for a temperature ≥40°C [5]. Although fever is a marker for bloodstream and other systemic infections, an association between fever and mortality has not been demonstrated in patients with sepsis or a BSI [17, 18]. A quick version of the PBS (qPitt) was recently derived based on 5 binary variables in the PBS, but excluded fever [18]. Hypothermia (temperature <36°C) was retained in the qPitt due to its independent association with mortality [18]. This temporal evolution has historical relevance, as temperature has been the most revised variable in the PBS during its first decade of existence (Table 1). The qPitt has higher discrimination in predicting mortality than the quick SOFA (qSOFA) in patients with Gram-negative BSIs [18]. An ongoing study is currently evaluating the qPitt in patients with Staphylococcus aureus BSIs (Battle SE, Justo J, Bookstaver PB, Kohn JE, Al-Hasan MN, personal communication).

Table 1.

Temporal Evolution of the Pitt Bacteremia Score

Authors (Year) [Reference]	Temperature (Point Allocation)	Other Variables (Point Allocation)	Critically Ill Definition
Hilf et al (1989) [1]	Not included	Hypotension, mechanical ventilation, coma	Any of the 3 criteria
Korvick et al (1991) [2]	>39°C (2)	Hypotension (2), mechanical ventilation (2), cardiac arrest (4); mental status: disorientation (1), stupor (2), coma (4)	Score ≥4 points
Chow et al (1991) [3]	37.6–39.9°C (1) ≥40°C (2)		Score ≥4 points
Korvick et al (1992) [4]	38.1–39.0°C (1) >39°C (2)		Score ≥4 points
Chow & Yu (1999) [5]	39.0-39.9°C (1) ≥40°C (2) 35.1–36.0°C (1) ≤35°C (2)		Pitt bacteremia score ≥4 points
Battle et al (2019) [18]	<36°C (1)	Hypotension (1), respiratory rate ≥25/minute or mechanical ventilation (1), cardiac arrest (1), altered mental status (1)	Quick Pitt bacteremia score (qPitt) ≥2 points

In the current issue of Clinical Infectious Diseases, Henderson and colleagues [19] evaluated both the PBS and the qPitt in a prospective, multicenter cohort of hospitalized patients with carbapemen-resistant Enterobacteriaceae (CRE) infections. Although the PBS has been previously used in patients with sepsis (with or without BSIs) [9], this is the first investigation to validate the PBS specifically in patients without BSIs. Overall, there were 475 patients included in the cohort from the Consortium on Resistance Against Carbapenems in Klebsiella and other Enterobacteriaceae (CRACKLE-1) database, including 351 (74%) without BSIs and 124 (26%) with BSIs. Temperature, as categorized in the PBS, was not associated with mortality. However, when dichotomized as in the qPitt, there was a 2-fold increase in mortality in patients with temperatures <36°C, as compared to those with temperatures ≥36°C. Both the PBS and the qPitt had high discrimination in predicting 14-day mortality in patients without BSIs (c-statistic 0.85 for both) [19]. The best cut-off points in the PBS and qPitt were ≥4 and ≥2, respectively, in patients without BSIs, similar to previously established break points for BSIs (Table 1).

Despite the limitations of the observational design and the temporal decline in mortality over the 4.5-year study period following the introduction of novel antimicrobial treatment options for CRE infections, the current study has several clinical and research implications. Most importantly, it validates the PBS and the qPitt in patients without BSIs. Since the respiratory and urinary tracts were the most common sources of CRE infections, the current results support the use of simple clinical scores for acute severity of illness (eg, the PBS and the qPitt) in future studies of pneumonia and complicated urinary tract infections. Moreover, by establishing cut-off points for the PBS and the qPitt in patients without BSIs, the study identifies patients with life-threatening infections who may benefit from future interventions, such as combination antimicrobial therapy, continuous infusion of beta-lactams, or other novel treatment strategies. This is particularly important for future clinical research designs, since survival benefits from appropriate antimicrobial therapy have not been demonstrated in patients with a PBS <4 or a qPitt <2, even in the presence of a BSI [13, 18]. Softer clinical endpoints than mortality, such as the recently proposed early clinical failure criteria, may be more reasonable in noncritically ill patients due to high overall survival rates [20].

The definition of sepsis as a life-threatening infection by the most recent sepsis-3 criteria has sparked a scientific debate regarding the most accurate acute severity of illness score [21]. Despite the aforementioned advantages of the PBS over other scores, it was overlooked in this most recent effort [22]. This was most likely due, in large part, to its name, which implied it was only applicable to patients with confirmed BSIs. However, none of the variables in the PBS are particularly specific to BSIs, and it is conceivable that it may be used in any patients with suspected infections. The ability to calculate a score at the bedside within minutes of a patient’s presentation to medical care and the high precision in the current and previous studies make the PBS and the qPitt certain candidates for the evaluation of patients with suspected sepsis. In particular, investigations are needed to define the utility of the PBS and the qPitt in identifying patients with life-threatening infections among those presenting to emergency departments and/or admitted to ICUs with suspected infections.

It is truly extraordinary for anything in medicine, whether a concept or a scoring system, to be as reliable, precise, and resilient as the PBS despite tremendous clinical, technological, and molecular advancements over the past 30 years. The current study, coupled with other contemporary work, solidifies the status of the PBS as the crown jewel of all acute severity of illness scores. Expanding the use of the PBS to infected patients without BSIs provides novel opportunities for future clinical investigations.

Note

Potential conflicts of interest. The authors: No reported conflicts of interest. Both authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.